jevyoumean
Suggest documented CLI subcommands by intent when spelling distance cannot capture what the user meant.
What Jev decides
The best candidate from names and help descriptions, or that no suggestion is reliable enough.
The jym wrapper detects an undocumented subcommand, extracts real alternatives and descriptions from help output, and asks Jev which candidate best matches the attempted command before the user chooses to run or cancel it.
- Jev role
- Choose which documented CLI subcommand most closely matches the intent of an unknown subcommand.
- Input
- The attempted subcommand and the documented candidate names and descriptions extracted from CLI help.
- Output
- A Choice distribution over documented alternatives.
- Decision frequency
- Only when a wrapped interactive command contains an undocumented subcommand.
- Fallback
- Ask before executing a suggestion, allow the original command or cancellation, and bypass correction for scripts, recursion, ambiguous flag values, or low confidence.
How it works
Overview
jevyoumean, invoked as jym, wraps selected command-line programs. It first checks the argument vector against the documented command tree. When the proposed subcommand is missing, it reads the target CLI help, presents only documented alternatives to Jev, and asks which meaning best fits. This can connect terms such as remove to rm or undo to restore even when edit distance is poor.
The original command is never run before correction. An interactive prompt shows the suggested command and probability, and the user can accept it, run the original input, or cancel. Scripts and non-interactive stderr pass through unchanged. The maintainer explicitly describes the project as experimental and not suitable for production-critical workflows.
- Language
- Go
- License
- MIT
- Author
- syumai
Sources
Related projects
Distill
Route coding-agent work, reasoning effort, and context retention through bounded Jev decisions.
What Jev decides
Which configured model and effort should handle a call, what evidence deserves attention, whether another attempt is useful, and what context should be retained.
jev-model-router
Route Claude Code model tiers and reasoning effort with one parallel Jev classification.
What Jev decides
Which fast, balanced, or deep tier fits the work, how much reasoning it needs, and whether production risk requires escalation.
jev-ultrafast
A browser agent where Jev selects an operation and a compatible page element from a live indexed action space.
What Jev decides
Which supported browser operation to run, and which compatible visible element to target